In depth map based rendering 3D video coding, a depth map is coded as a texture view picture into a coded bitstream. However, depth map data show signal characteristics different from natural video data. The most intuitive one is that a depth map includes a lot of smooth areas with similar sample values. For most of the cases, the samples within the smooth area share a same depth value.
In high efficiency video coding (HEVC)-based depth coding, a coding mode termed as a single depth mode is used to efficiently code smooth areas within a picture. The concept of the single depth mode is to simply reconstruct a current coding unit (CU) as a smooth area with a single depth sample value. Thus a smooth area within a depth map may be coded more efficiently through more concise syntax.
To reconstruct a CU coded with the single depth mode, a sample candidate list is constructed by inserting into the candidate list sample candidates derived from neighboring samples of a current CU in a predefined order. Complicated processing is performed to select a best candidate from the candidate list to fill the current CU.